The broad goal of the work in this application is to improve the care of hospitalized older adults through safer medication prescribing practices. More specifically, the objectives of this proposal are to define current usage patterns and risks of sedating medications in hospitalized patients, and use this knowledge to develop an intervention to promote safer prescribing. This work will be carried out in two phases. First, two large existing hospital data repositories will be used in a parallel and complementary fashion to investigate current usage patterns of sedating medications and their association with specific hypothesized risks - hospital-acquired pneumonia and in-hospital falls - in two large cohorts of hospitalized patients. These analyses will employ advanced epidemiologic methods for confounder control, including propensity score and instrumental variable approaches. In the second phase of this work, the information gained from the first phase will be combined with existing knowledge related to safe prescribing practices to develop and conduct a pilot study of a computerized decision support (CDS) tool that bundles prescribing suggestions with specific preventive interventions for those at highest risk. The impact of this CDS tool on safe prescribing practices will be investigated, and this information will be used to inform development of a multi-site, cluster randomized controlled trial as part of a future R01 submission. Through these analyses, her planned coursework, and strong mentorship, the candidate, Dr. Shoshana Herzig, will advance her methodological skills in pharmacoepidemiologic analysis and aging research. She will also further develop her ability to translate research findings into interventions that will successfully impact the clinical care of older hospitalized patients, and learn the methods necessary to rigorously measure that impact. Taken together, this multi-faceted proposal will facilitate Dr. Herzig's development into an independent patient-oriented investigator in geriatric pharmacoepidemiology.